Nikon D800; it's going to be fascinating

"David J Taylor" wrote in
:

"Rich" wrote in message
...
[]
Only a problem is loss of resolution, which IS visible, even with a
D200
as conversion of said camera proved that. I want to see the results
of
this new AA-less camera versus the AA camera with a good lens. For
those
terrified of the potential for interference with fine, repeating
details,
they should stick with the AA filter.

Noticeable resolution drop is perhaps not surprising on a 10 MP camera,
but I thought we were talking nearer 36 MP?

David



Remains to be seen. If it's true they are going to be offering it, I
hope Nikon is offering it for more reasons than just marketing.

"Rich" wrote in message
...
[]
Remains to be seen. If it's true they are going to be offering it, I
hope Nikon is offering it for more reasons than just marketing.

It might make more sense to offer two different strengths of AA filter,
although my guess is that they will just offer one.

David

David J Taylor wrote:


- if you use lenses with no significant response above the Nyquist
frequency, then there's nothing for the AA filter to remove. I.e.
with lenses which aren't the sharpest it doesn't matter. Put another
way, if the camera is 36 MP but the lens only 18 MP, there may be no
need for AA filtering.

Very interesting.

Does this mean (changing the subject to compact small-sensor cameras) that
there may actually be a point to all this pixel cramming -- things like
putting 16 megapixels on a 1/2.3" sensor?

I'm assuming that such cameras have been using AA filters all along too.

"Neil Harrington" wrote in message
...
[]
Very interesting.

Does this mean (changing the subject to compact small-sensor cameras)
that there may actually be a point to all this pixel cramming -- things
like putting 16 megapixels on a 1/2.3" sensor?

I'm assuming that such cameras have been using AA filters all along too.

Having a fixed lens, where you can then design the lens and the sensor as
an integrated system, may well allow you to get better results than having
a "general purpose" sensor onto which a wide range of lens types can be
installed. It could well be that limiting the MTF of the lens not to
exceed the sensor needs is one design option.

In the limit, if you have a fixed sensor size, and a fixed display size,
how does the pixel density affect things? In the past smaller pixels have
meant a poorer light collection efficiency, and hence a poorer overall
signal to noise ratio. If that's fixed with micro-lenses, then the pixel
size primarily affects system MTF, and that is becoming more and more
lens-limited. Not to mention diffraction limited for the small-sensor
camera.

It raises the question as to what is the resolution requirement of the
final image. An 8 x 10 inch print at 250 ppi? 5 MP. So could the eye
actually see a significant difference between a 5 MP and a 15 MP sensor
(if they are equally efficient)? And might the 16 MP sensor not require
an AA filter as the lens doesn't justify it? And might that save cost,
and make the camera /sound/ better?

Having thought it through, no, I'm not convinced that the saving of cost
by not having an AA filter would alone justify the move to high-MP
sensors, but it might be a side-benefit, particularly on cheaper cameras.

Just some initial ideas - your thoughts welcomed.

Cheers,
David

"David J Taylor" wrote:
"Neil Harrington" wrote:
[]
Very interesting.

Does this mean (changing the subject to compact
small-sensor cameras) that there may actually be a
point to all this pixel cramming -- things like
putting 16 megapixels on a 1/2.3" sensor?

I'm assuming that such cameras have been using AA filters all along too.

Having a fixed lens, where you can then design the lens and the sensor as
an integrated system, may well allow you to get better results than having
a "general purpose" sensor onto which a wide range of lens types can be
installed. It could well be that limiting the MTF of the lens not to
exceed the sensor needs is one design option.

In the limit, if you have a fixed sensor size, and a fixed display size,
how does the pixel density affect things? In the past smaller pixels have
meant a poorer light collection efficiency, and hence a poorer overall
signal to noise ratio. If that's fixed with micro-lenses, then the pixel
size primarily affects system MTF, and that is becoming more and more
lens-limited. Not to mention diffraction limited for the small-sensor
camera.

It raises the question as to what is the resolution requirement of the
final image. An 8 x 10 inch print at 250 ppi? 5 MP. So could the eye
actually see a significant difference between a 5 MP and a 15 MP sensor
(if they are equally efficient)? And might the 16 MP sensor not require
an AA filter as the lens doesn't justify it? And might that save cost,
and make the camera /sound/ better?

Having thought it through, no, I'm not convinced that the saving of cost
by not having an AA filter would alone justify the move to high-MP
sensors, but it might be a side-benefit, particularly on cheaper cameras.

Just some initial ideas - your thoughts welcomed.

You both should go back and review David Littleboy's
comments about the difference between the shape factor
for an AA filter than the gentle rolloff of a lens.

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

"Floyd L. Davidson" wrote:

You both should go back and review David Littleboy's
comments about the difference between the shape factor
for an AA filter than the gentle rolloff of a lens.

ROFL. I got tired of yelling at them.

--
David J. Littleboy
Tokyo, Japan

"David J. Littleboy" wrote in message
...

"Floyd L. Davidson" wrote:

You both should go back and review David Littleboy's
comments about the difference between the shape factor
for an AA filter than the gentle rolloff of a lens.

ROFL. I got tired of yelling at them.

--
David J. Littleboy
Tokyo, Japan

But the AA filter doesn't have a sharp cut-off either, so what really
matters is the system MTF above Nyquist. You need to decide what MTF
matters - whether it's 5%, 2%, 1% or whatever. It's a compromise. So it
would not be surprising if, with some lenses and a high resolution sensor,
the sensor resolution was sufficiently higher than than that of the lens
that the AA filter was doing virtually nothing.

Don't get me wrong - I'm not the one advocating omitting the AA filter,
but I can see circumstances where it may not be required.

Cheers,
David T

"David J. Littleboy" wrote:
"Floyd L. Davidson" wrote:

You both should go back and review David Littleboy's
comments about the difference between the shape factor
for an AA filter than the gentle rolloff of a lens.

ROFL. I got tired of yelling at them.

I noticed. :-)

You did a pretty darn good job of describing why it is
as it is, but... Maybe it would take a couple graphs,
showing the relationship between a brick wall and a grassy
slope.

HIGH QUALITY LENS
Nyquist
| Area 'A' is desireable detail
---|---.|,|, that should not be lost
| "|"---...
| "'- Area 'B' is detail that is sacrificed
| \____ to avoid aliasing
---+----+----+-----------
A B C Area 'C' is detail above Nyquist
Resolution -- that causes aliasing



DECENT AA FILTER
Nyquist
|
---|-, | This filter loses area above
*. | the line at B, but totally
|,| eliminates signal above at C
*._|____
---+----+----+-----------
A B C
Resolution --



POOR QUALITY LENS
Nyquist
|
-. |
*|. | The lens loses data above
''|,, the line at A, and does not
|--| eliminate signal above the
| ''--..___ line at C.
---+----+----+-----------
A B C
Resolution --

The bottom line is that using the lens as an AA filter
will result in a low resolution image that also has
effects from aliasing. Or, "What's the point?"

Whatever, I'll be just really amazed if Nikon has a model
without an AA filter.

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

"David J Taylor" wrote:
But the AA filter doesn't have a sharp cut-off either,

But it is designed to have the sharpest cutoff that can
be economically produced with existing technology.

A lens on the other hand, even a really poor example
that exhibits low resolution, is specifically designed
to *not* have a sharp cutoff.

The rest of this rationalization is nonsense:

so what really
matters is the system MTF above Nyquist. You need to decide what MTF
matters - whether it's 5%, 2%, 1% or whatever. It's a compromise. So it
would not be surprising if, with some lenses and a high resolution sensor,
the sensor resolution was sufficiently higher than than that of the lens
that the AA filter was doing virtually nothing.

The AA filter is compromised/optimized in one direction
and the lens is optimized/compromised in the other
direction, and you are saying that since they aren't
exactly perfect that they might be the same, which just
isn't true.

If the AA filter would do "virtually nothing", then why
go to the expense of a high resolution sensor? It's a
two bit lens and a two bit sensor will be quite good
enough for the 50 cent results.

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

The AA filter is compromised/optimized in one direction
and the lens is optimized/compromised in the other
direction, and you are saying that since they aren't
exactly perfect that they might be the same, which just
isn't true.

No. I'm saying that if the MTF of the lens is small at the Nyquist
frequency, there is little for the AA filter to do as there will be only a
small amount of high frequency information present.

If the AA filter would do "virtually nothing", then why
go to the expense of a high resolution sensor? It's a
two bit lens and a two bit sensor will be quite good
enough for the 50 cent results.

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

Indeed - why do people need 36 MP sensors in DSLRs, or 16 MP sensors in
consumer cameras where the prints won't be bigger than 6 x 4 inches (or
whatever)? You may be stuck with a higher sensor resolution than you need
just because you have bought the camera for its other features.

With high-resolution sensors, and good quality lenses, I would not
recommend using a camera without an AA filter.

David